Parser&Walker: support literals in regular exp

6 files changed, 25 insertions, 13 deletions

diff --git a/bashast/bashast.g b/bashast/bashast.g
index 92f4f92..c6d7f34 100644
--- a/bashast/bashast.g
+++ b/bashast/bashast.g
@@ -80,6 +80,7 @@ tokens{
 	BRANCH;
 	MATCH_PATTERN;
 	MATCH_REGULAR_EXPRESSION;
+	ESCAPED_CHAR;
 	NOT_MATCH_PATTERN;
 	MATCH_ANY;
 	MATCH_ANY_EXCEPT;
@@ -651,13 +652,20 @@ scope {
 }
 	:(
 		DQUOTE! { $bash_pattern_part::quoted = !$bash_pattern_part::quoted; }
-		|	{$bash_pattern_part::quoted}? => ~DQUOTE
+		|	{$bash_pattern_part::quoted}? => preserved_tokens
 		|	(ESC BLANK) => ESC BLANK
 		|	LPAREN { if(LA(-2) != ESC) $bash_pattern_part::parens++; }
 		|	LLPAREN { if(LA(-2) != ESC) $bash_pattern_part::parens += 2; }
 		|	{$bash_pattern_part::parens != 0}? => RPAREN { if(LA(-2) != ESC) $bash_pattern_part::parens--; }
 		|	~(BLANK|EOL|LOGICAND|LOGICOR|LPAREN|RPAREN|DQUOTE|LLPAREN)
 	 )+;
+
+preserved_tokens
+	:	non_dquote -> ESCAPED_CHAR non_dquote;
+
+non_dquote
+	:	~DQUOTE;
+
 keyword_binary_string_operator
 	:	BLANK! binary_operator BLANK!
 	|	BLANK! EQUALS BLANK!
diff --git a/bashast/gunit/cond_main.gunit b/bashast/gunit/cond_main.gunit
index cb8ffef..b6f339b 100644
--- a/bashast/gunit/cond_main.gunit
+++ b/bashast/gunit/cond_main.gunit
@@ -31,6 +31,7 @@ condition_expr:
 "[ a == b ]" -> (BUILTIN_TEST (= (STRING a) (STRING b)))
 "[ a != b ]" -> (BUILTIN_TEST (NOT_EQUALS (STRING a) (STRING b)))
 "[[ \"${DISTUTILS_SRC_TEST}\" =~ ^(setup\.py|nosetests|py\.test|trial(\ .*)?)$ ]]" -> (KEYWORD_TEST (MATCH_REGULAR_EXPRESSION (STRING (DOUBLE_QUOTED_STRING (VAR_REF DISTUTILS_SRC_TEST))) (STRING ^ ( setup \ . py | nosetests | py \ . test | trial ( \   . * ) ? ) $)))
+"[[ a =~ \" \"bcd ]]" -> (KEYWORD_TEST (MATCH_REGULAR_EXPRESSION (STRING a) (STRING ESCAPED_CHAR   bcd)))
 "[ -n \"$FROM_LANG\" -a -n \"$TO_LANG\" ]" -> (BUILTIN_TEST (BUILTIN_LOGIC_AND (n (STRING (DOUBLE_QUOTED_STRING (VAR_REF FROM_LANG)))) (n (STRING (DOUBLE_QUOTED_STRING (VAR_REF TO_LANG))))))
 "[ -n \"$FROM_LANG\" -o -n \"$TO_LANG\" ]" -> (BUILTIN_TEST (BUILTIN_LOGIC_OR (n (STRING (DOUBLE_QUOTED_STRING (VAR_REF FROM_LANG)))) (n (STRING (DOUBLE_QUOTED_STRING (VAR_REF TO_LANG))))))
 "[ -n \"a\" -o -n \"a\" -a -n \"a\" ]" -> (BUILTIN_TEST (BUILTIN_LOGIC_OR (n (STRING (DOUBLE_QUOTED_STRING a))) (BUILTIN_LOGIC_AND (n (STRING (DOUBLE_QUOTED_STRING a))) (n (STRING (DOUBLE_QUOTED_STRING a))))))
diff --git a/bashast/libbashWalker.g b/bashast/libbashWalker.g
index d45d9c3..f57f403 100644
--- a/bashast/libbashWalker.g
+++ b/bashast/libbashWalker.g
@@ -302,6 +302,7 @@ string_part returns[std::string libbash_value, bool quoted, bool is_raw_string]
 		$libbash_value = transformed.str();
 		$quoted = true;
 	}
+	|(ESCAPED_CHAR) => ESCAPED_CHAR libbash_string=any_string { $libbash_value = "\\" + libbash_string; }
 	|(libbash_string=any_string {
 		$libbash_value = libbash_string;
 	});
@@ -797,7 +798,7 @@ keyword_condition returns[bool status]
 	} r=keyword_condition) { $status= l && r; }
 	|^(NEGATION l=keyword_condition) { $status = !l; }
 	|^(MATCH_REGULAR_EXPRESSION left_str=string_expr right_str=string_expr) {
-		bash_ast ast(std::stringstream(right_str.libbash_value), &bash_ast::parser_all_expansions);
+		bash_ast ast(std::stringstream(right_str.libbash_value), &bash_ast::parser_all_expansions, false);
 		std::string pattern = ast.interpret_with(*walker, &bash_ast::walker_string_expr);
 		boost::xpressive::sregex re = boost::xpressive::sregex::compile(pattern);
 		$status = boost::xpressive::regex_match(left_str.libbash_value, re);
diff --git a/scripts/var_expansion.bash b/scripts/var_expansion.bash
index 4bdcb3e..97e9587 100644
--- a/scripts/var_expansion.bash
+++ b/scripts/var_expansion.bash
@@ -135,8 +135,6 @@ root=123
 echo "${search_paths/%/${root}}"
 echo "${search_paths/#/${root}}"
 
-# This regular expression will cause boost::exception_detail::clone_impl<boost::xpressive::regex_error>
-#[[ "${version_components_groups}" =~ ("*".*" "|" *"|^2.*\ (2|\*)|^3.*\ (3|\*)) ]]
 [[ " ${FUNCNAME[@]:2} " =~ " "(_python_final_sanity_checks|python_execute_function|python_mod_optimize|python_mod_cleanup)" " ]]
 [[ "$(declare -p PYTHON_SANITY_CHECKS_EXECUTED)" != "declare -- PYTHON_SANITY_CHECKS_EXECUTED="* || " ${FUNCNAME[@]:1} " =~ " "(python_set_active_version|python_pkg_setup)" " && -z "${PYTHON_SKIP_SANITY_CHECKS}" ]]
 [[ " ${FUNCNAME[@]:1} " =~ " "(python_set_active_version|python_pkg_setup)" " ]]
@@ -146,3 +144,4 @@ PYTHON_DEPEND="2:2.6"
 version_components_group_regex="(2|3|\*)(:([[:digit:]]+\.[[:digit:]]+)?(:([[:digit:]]+\.[[:digit:]]+)?)?)?"
 version_components_groups="${PYTHON_DEPEND}"
 [[ "${version_components_groups}" =~ ^((\!)?[[:alnum:]_-]+\?\ )?${version_components_group_regex}(\ ${version_components_group_regex})?$ ]] && echo true
+[[ "${version_components_groups}" =~ ("*".*" "|" *"|^2.*\ (2|\*)|^3.*\ (3|\*)) ]] && echo true
diff --git a/src/core/bash_ast.cpp b/src/core/bash_ast.cpp
index 5cca4f4..23571ef 100644
--- a/src/core/bash_ast.cpp
+++ b/src/core/bash_ast.cpp
@@ -36,31 +36,34 @@
 #include "libbashParser.h"
 #include "libbashWalker.h"
 
-void bash_ast::read_script(const std::istream& source)
+void bash_ast::read_script(const std::istream& source, bool trim)
 {
   std::stringstream stream;
   stream << source.rdbuf();
   script = stream.str();
   boost::algorithm::erase_all(script, "\\\n");
-  boost::trim_if(script, boost::is_any_of(" \t\n"));
+  if(trim)
+    boost::trim_if(script, boost::is_any_of(" \t\n"));
 }
 
 bash_ast::bash_ast(const std::istream& source,
-                   std::function<pANTLR3_BASE_TREE(plibbashParser)> p): parse(p)
+                   std::function<pANTLR3_BASE_TREE(plibbashParser)> p,
+                   bool trim): parse(p)
 {
-  read_script(source);
+  read_script(source, trim);
   init_parser("unknown source");
 }
 
 bash_ast::bash_ast(const std::string& script_path,
-                   std::function<pANTLR3_BASE_TREE(plibbashParser)> p): parse(p)
+                   std::function<pANTLR3_BASE_TREE(plibbashParser)> p,
+                   bool trim): parse(p)
 {
   std::stringstream stream;
   std::ifstream file_stream(script_path);
   if(!file_stream)
     throw libbash::parse_exception(script_path + " can't be read");
 
-  read_script(file_stream);
+  read_script(file_stream, trim);
   init_parser(script_path);
 }
 
diff --git a/src/core/bash_ast.h b/src/core/bash_ast.h
index 50ca7a6..073553c 100644
--- a/src/core/bash_ast.h
+++ b/src/core/bash_ast.h
@@ -67,7 +67,7 @@ class bash_ast: public boost::noncopyable
 
   typedef std::unique_ptr<libbashWalker_Ctx_struct, std::function<void(libbashWalker_Ctx_struct*)>> walker_pointer;
 
-  void read_script(const std::istream& source);
+  void read_script(const std::istream& source, bool trim);
   void init_parser(const std::string& script_path);
   walker_pointer create_walker(interpreter& walker,
                                antlr_pointer<ANTLR3_COMMON_TREE_NODE_STREAM_struct>& nodes);
@@ -77,13 +77,13 @@ public:
   /// \param source input source
   /// \param p the parser rule for building the AST
   bash_ast(const std::istream& source,
-           std::function<pANTLR3_BASE_TREE(libbashParser_Ctx_struct*)> p=parser_start);
+           std::function<pANTLR3_BASE_TREE(libbashParser_Ctx_struct*)> p=parser_start, bool trim=true);
 
   /// \brief build AST from string
   /// \param script_path input source
   /// \param p the parser rule for building the AST
   bash_ast(const std::string& script_path,
-           std::function<pANTLR3_BASE_TREE(libbashParser_Ctx_struct*)> p=parser_start);
+           std::function<pANTLR3_BASE_TREE(libbashParser_Ctx_struct*)> p=parser_start, bool trim=true);
 
   /// \brief the functor for walker start rule
   /// \param tree_parser the pointer to the tree_parser